Wijeysundera et al1 should be commended for their thought-provoking article suggesting that preoperative medical consultation is associated with worse outcomes. They performed a detailed analysis of administrative data, attempted to minimize bias with appropriate statistical techniques, and found a positive association between preoperative consultation and mortality.
The results, however compelling, are likely a product of confounding by indication, which is a significant problem when using administrative databases. Their propensity score analysis and subgroup analyses served to minimize such bias by equalizing the groups, but statistics alone cannot eliminate it. The proportion of patients with coronary artery disease, diabetes, congestive heart failure, and peripheral vascular disease was equal between the groups because of matching. But what about the severity and number of the underlying comorbidities? The “brittle” diabetic patient is more likely to be referred than the diet-controlled diabetic patient with normoglycemia. The hypertensive patient with a systolic blood pressure of 200 mm Hg is more likely to be referred than the patient with a pressure of 145 mm Hg. Among all of the comorbidities, the sickest and therefore most likely to die are also the most likely to be referred. Propensity matching in an administrative database cannot eliminate this.